Positron emission tomography (PET) and magnetic resonance imaging (MRI) are routinely utilized in cervical cancer management. Their comparative effectiveness has not been well defined. These advanced and costly modalities are not used clinically for assessment of tumor heterogeneity, known to critically influence individual patients' treatment outcome. Cervical cancer is the 3rd leading cause of cancer death in women worldwide and one third of patients fail treatment. If treatment failure can be predicted early, potential more intense therapy options are available to substantially improve ultimate outcome. However, current established clinical criteria lack the sensitivity to predict ultimate treatment failure for each individual patient. Clinical MRI and PET can obtain functional imaging biomarkers for each tumor voxel, but have not been clinically applied to characterize tumor heterogeneity in cervix cancer. Our preliminary results show that functional tumor heterogeneity assessed by MRI and PET has strong potential to improve personalized, therapy-specific, and early outcome prediction. We developed a novel Voxel-Histogram (VH) method readily characterizing spatial and temporal variation of the tumor voxels' functional heterogeneity. We established functional imaging metrics to quantify tumor voxels at risk for therapy failure based on unfavorable imaging biomarkers values: high 8F- fluorodeoxyglucose uptake (SUV), low Dynamic Contrast Enhancement (DCE) and low Apparent Diffusion Coefficient (ADC). The heterogeneity imaging metrics include biomarkers' VH, optimal thresholds to define at-risk tumor voxels and absolute functional risk volume (AFRV), were derived from at-risk tumor voxels for early outcome prediction. We hypothesize that the ability to quantify at-risk tumor voxels substantially improves the timing and power for early outcome prediction. We found that heterogeneity DCE metrics predict, as early as 2 weeks into treatment, the ultimate therapy outcome. The predictive power was further improved by integrating heterogeneity imaging metrics with the routinely available clinical prognosticators. The purpose of the current proposal is to characterize and quantify tumor heterogeneity with DCE, ADC and FDG metrics at various treatment times in cervix cancer patients, and to correlate these imaging metrics with tumor recurrence and cancer death. We have 3 specific aims: 1) Refine and advance functional imaging-based Voxel-Histogram and imaging metrics to characterize tumor heterogeneity and quantify at-risk tumor voxels, 2) Validate the predictive power of MRI and PET imaging metrics with treatment outcome and rank the comparative effectiveness at different time points, and 3) Derive novel multivariate predictive algorithms from imaging metrics and established clinical prognosticators to further enhance predictive power. Upon completion, tumor-heterogeneity based functional imaging algorithms will be available to define the best, earliest and most effective imaging metrics and time points for PET and/or MRI to maximally impact patients' treatment strategy.